[Mlir-commits] [mlir] [mlir][sparse] Populate lvlToDim (PR #68937)

[Aart Bik]

================
@@ -48,15 +48,17 @@ bool mlirAttributeIsASparseTensorEncodingAttr(MlirAttribute attr) {
 MlirAttribute
 mlirSparseTensorEncodingAttrGet(MlirContext ctx, intptr_t lvlRank,
                                 MlirSparseTensorDimLevelType const *lvlTypes,
-                                MlirAffineMap dimToLvl, int posWidth,
-                                int crdWidth) {
+                                MlirAffineMap dimToLvl, MlirAffineMap lvlToDim,
+                                int posWidth, int crdWidth) {
   SmallVector<DimLevelType> cppLvlTypes;
   cppLvlTypes.reserve(lvlRank);
   for (intptr_t l = 0; l < lvlRank; ++l)
     cppLvlTypes.push_back(static_cast<DimLevelType>(lvlTypes[l]));
-  mlir::AffineMap lvlToDim; // TODO: provide in API
+  auto unwrappedLvlToDim = unwrap(lvlToDim);
+  if (!unwrappedLvlToDim)
+    unwrappedLvlToDim = inferLvlToDim(unwrap(dimToLvl), unwrap(ctx));
----------------
aartbik wrote:

I am not sure if we should infer inside the CAPI construction (which is really a binding to the outside world), or whether this should just mechanically accept whatever is given. If we do inference during construction, if feels this logic should be in SparseTensorEncodingAttr::get itself, not in the binding code



https://github.com/llvm/llvm-project/pull/68937